Head Pose Estimation on Low Resolution Images
|
|
- Berenice Norton
- 8 years ago
- Views:
Transcription
1 Head Pose Estimation on Low Resolution Images Nicolas Gourier, Jérôme Maisonnasse, Daniela Hall, James L. Crowley PRIMA, GRAVIR-IMAG INRIA Rhône-Alpes, St. Ismier. France. Abstract. This paper addresses the problem of estimating head pose over a wide range of angles from low-resolution images. Faces are detected using chrominance-based features. Grey-level normalized face imagettes serve as input for linear auto-associative memory. One memory is computed for each pose using a Widrow-Hoff learning rule. Head pose is classified with a winner-takes-all process. We compare results from our method with abilities of human subjects to estimate head pose from the same data set. Our method achieves similar results in estimating orientation in tilt (head nodding) angle, and higher precision for estimating orientation in the pan (side-to-side) angle. 1. Introduction Knowing the head pose of a person provides important cues concerning visual focus of attention [12]. Applications such as video surveillance, intelligent environments and human interaction modelling require head pose estimation from low-resolution face images. Unfortunately, most methods described in the research literature require high-resolution images, often using multiple views of the face. In this paper we address the problem of estimating head pose from low-resolution single images. The pose, or orientation, of a head is determined by 3 angles: slant, pan and tilt. The slant angle represents the person s head inclination with regard to the image plane, whereas the tilt and the pan angles represent the vertical and the horizontal inclination of the face. Our objective is to obtain a reliable estimation of head pose on unconstrained low-resolution images. We employ a fast, chrominance-based segmentation algorithm to isolate and normalize the face region in size and slant. We then project this region of the image into a small fixed-size imagette using a tranformation that normalises size and slant orientation. Normalised face imagettes are used to train an auto-associative memory using the Widrow-Hoff correction rule. Classification of head pose is obtained by comparing normalised face imagettes with those reconstructed by the auto-associative memory. The head pose which obtains the highest score is selected. We compare results with our method to human performance on head pose estimation using the same data set [13]. This process is described in section 3. We compare results from
2 2 this method with human performance for head pose estimation using the same data set, as described in section 4. Results of these comparisons are discussed in section Approaches to Head Pose Estimation Local or global approaches exist for head pose estimation. Local approaches usually estimate head pose from a set of facial features such as eyes, eyebrows and lips. Three dimensional rotation of the head can be estimated from correspondences between such facial landmarks in the image and the face [1], [2], [3]. However, the detection of facial features tends to be sensitive to partial changes of illumination, person and pose variations. Robust techniques have been proposed to handle such variations [4], [5] but these require high resolution images of the face and tracking can fail when certain facial features are occluded. Some local-based systems, such as FaceLAB [21], have a precision smaller than one degree. Such systems use stereo vision and require high resolution of the image of the face. Transformation-based approaches use some geometric properties of facial landmarks to estimate the 3D rotation of the head [6], [7], [8]. However, such techniques remain sensitive to the precision of detected regions and to the resolution of the face image. Such problems do not appear when using global approaches. Global approaches use the entire image of the face to estimate head pose. The principal advantage of global approaches is that only the face needs to be located. No facial landmark, or face model are required. Global approaches can accommodate very low resolution images of the face. Template matching is a popular method to estimate head pose. The best template is found via a nearest-neighbour algorithm, and the pose associated with this template is selected as the best pose. Template matching can be performed using Gabor Wavelets and Principle Components Analysis (PCA) [9], or Support Vector Machines [10], but these approaches tend to be sensitive to alignment and are dependent on the identity of the person. Neural networks have also been used for head pose estimation [11]. Stiefelhagen [12] reports 10 degrees of precision on the Pointing 04 Head Pose Image Database [13]. However, some images of the same users were used both in training and testing. Furthermore, the number of cells in hidden layers is chosen arbitrarily, which prevent creation of image class prototypes. In the method described in this paper, we adapt auto-associative memories based on the Widrow-Hoff learning rule. Auto-associative memories require very few parameters and contain no hidden layers [14]. Prototypes of image classes can be saved and reused. The Widrow-Hoff learning rule provides robustness to partial occlusions [22]. Each head pose serves to train an auto-associative network. Head pose is estimated by selecting the auto-associative network with the highest likelihood score.
3 3 3. Head Pose Estimation using Linear Auto-associative Neural Networks 3.1 Linear Auto-associative Memories Linear auto-associative memories are a particular case of one-layer linear neural networks where input patterns are associated with each other. Auto-associative memories associate images with their respective class, even when the image has been degraded or partially occluded. With this approach, each cell corresponds to an input pattern. We describe a grey-level input image x with a normalized vector x = x / x. A set of M images composed of N pixels of the same class are stored into a N x M matrix X = (x 1, x 2,, x M ). The linear auto-associative memory is represented by a connection matrix W. The reconstructed image y k is obtained by computing the product between the source image x and the connection weighted matrix W k : y k = W k x. The similarity between the source image and a class k of images is estimated as the cosine between x and y k : cos(x, y) = x y T. A similarity of 1 corresponds to a perfect match. The connection matrix W k 0 is initialized with the standard Hebbian learning rule W k 0 = X k X k T. Reconstructed images with Hebbian learning are equal to the first eigenface of the image class. To improve recognition abilities of the neural network, we learn W with the Widrow-Hoff rule. 3.2 The Widrow-Hoff Correction Rule The Widrow-Hoff correction rule is a local supervised learning rule. At each presentation of an image, each cell modifies its weights from the others. Images X, of the same class are presented iteratively with an adaptation step η until all images are classified correctly. As a result, the connection matrix W k becomes spherically normalized. The Widrow-Hoff learning rule is described by: W t+ 1 k = W t k + ( x W x) x t T η k (1) In-class images are minimally deformed by multiplying with the connection matrix, while extra-class images are more strongly deformed. Direct comparison between input and output normalized images gives a score between 0 and 1. This correction rule has shown good results on classic face analysis problems in the case of images from a single camera, for problems such as face recognition, sex classification and facial type classification [14]. The Widrow-Hoff correction rule increases the performance of PCA and provides robustness to partial occlusions [22]. All dimensions are used and few parameters are needed. There is no requirement to specify the choice of the structure or the number of cells in hidden layers. Furthermore, prototypes, W k, of image classes can be saved, recovered and directly reused on other images unlike non-linear
4 4 memories or neural networks with hidden layers, where prototypes can not be recovered. 3.3 Head Pose Image Database The choice of a good database is crucial for learning. The Pointing 04 Head Pose Image database [13] consists of 15 sets of images of different people. Each set contains 2 series of 93 images of the same person at different poses. Subjects are 20 to 40 years old. Five people have facial hair and seven are wearing glasses. Figure 1. Sample of the Pointing 04 Image Database Head pose is determined by pan and tilt angle. Each angle varies between -90 and +90 degrees, with a step of 15 degrees for pan, and 30 and 15 for tilt. Negative values for tilt correspond to bottom poses and positive values correspond to top poses. During the database acquisition, people were asked to look successively at 93 markers. Each marker corresponds to a particular pose. A sample of the database can be seen Figure Head Pose Prototypes The face region is normalized into a low resolution normalized grey-scale imagette of 23x30 pixels, as in [4]. Face normalization provides invariance of position, scale and slant [15]. This increases the reliability of results and processing time becomes independent of original face size. All further operations take place within this imagette. We consider each head pose as a class. A connection matrix W k is computed for each pose k. The Pointing 04 database consists in 13 Poses for pan and 9 Poses for tilt. Two experiments have been performed using this approach. In the separate technique, we learn each angle on an axis while varying the angle of the other axis. Each classifier corresponding to a pan angle is trained with varying tilt angle. Similarly, each memory corresponding to a tilt angle is trained with a varying pan angle. The separate experiment learns 22 classifiers: Pan = +90,, Pan = -90, Tilt = +90,, Tilt = -90. We use an adaptation step η of for pan and for tilt for this experiment. Pan and tilt are trained separately.
5 5 In the grouped experiment, pan and tilt angle are trained together. Each classifier corresponds to a pan and a tilt angle. This experiment learns 93 classifiers: (Pan,Tilt)=(0,-90),, (Pan, Tilt)=(+90,+75), (Pan, Tilt)=(0,+90). We use an adaptation step η of for this experiment. To estimate head pose on a given face imagette, a simple winner-takes-all process is employed. We compute the cosine between the source image X and reconstructed images X k. The pose whose memory obtains the best match is selected (2). Pose = arg max(cos(x, X k )) (2) k 4. Human Abilities for Head Pose Estimation To our knowledge, there is no data avalaible concerning human abilities for estimating head pose from images. Kersten [18] reports that front and profile poses are particularly well recognized by humans. These poses are used as key poses [19]. This observation is true not only for head poses, but also for other objects. However, they do not estimate intermediate poses. As a comparison to our artificial system, we measured the performance of a group of 72 human subjects on head pose estimation. In our experiment, we have tested 36 men and 36 women, ranging in age from 15 to 60 years old. The experiment consisted of two parts: one for pan angle estimation, and the other for tilt angle. Images from the Pointing 04 Head Pose Database were presented in random order to the subject for 7 seconds, with a different order for each subject. Subjects were asked to examine the image, and to select an answer pose estimation from a fixed set. The data base consists of 65 images for pan and 45 for tilt, which gives 5 images for each pose. The psycho-physical basis for human head pose estimation from static images is unknown. We do not know whether humans have a natural ability to estimate head pose from such images, or whether people must be trained for this task using annotated images. In order to avoid bias in our experiment, the subjects were divided into 2 groups: people in the first group may inspect the labelled training images of head pose as long as they wish before beginning the experiment, whereas people in the second group are not provided an opportunity to see the images before the experiment. First and second groups are respectively referred as Calibrated and Non-Calibrated subjects. Creating these two groups allows us to determine if training significantly increases human performances on head pose estimation. 5. Results and Discussion In this section, we compare results of the two variations of our method (separate and grouped) using the Pointing 04 Head Pose image database. There are two ways of splitting the data for training and testing. By using the first set of the database as the training data and testing on the second set, we measure the performance of our system on known users. By using the Jack-Knife method, also known as the leave-one-out
6 6 algorithm, we measure the performance on unknown users. To have an idea of the efficiency of our system in human-computer interaction applications, we compare performances of our system with human performances. 5.1 Evaluation Measures To evaluate the performance of our system, we must define evaluation criteria. Average absolute error for pan and tilt is the main evaluation metric. It is computed by averaging the difference between expected pose and estimated pose for all images. We also compute average absolute error for pan and tilt per pose. The Pointing 04 database is well suited for such measure, because it provides the same amount of data for each pose. Precise classification rate and correct classification with 15 degrees errors is also computed. We compare results of our system on known and unknown users. Results are presented in Table Performances Our system works well with known subjects on both angles. With the separate technique, we achieve a mean error of 7.3 degrees in pan and 12.1 degrees in tilt. The grouped learning provides a mean error of 8.5 degrees in pan and 10.1 degrees in tilt. Pan angle can be correctly estimated with a precision of 15 degrees in more than 90% of cases with both learning techniques. Results obtained with the Jack-Knife algorithm show that our system also generalizes well to previous unseen subjects and is robust to identity. With the separate technique, we see that pan angle is well recognized with an average error of 10.3 degrees. Average error decreases to 10.1 degrees using the grouped learning. The average tilt error is 15.9 degrees using the separate technique, and 16.8 degrees using the grouped technique. Average error per pose is shown in Figure 2. Known Users LAAM separate LAAM grouped Pan Average Error 7.3 º 8.5 º Tilt Average Error 12.1 º 10.1 º Pan Class. With 0º 61.3 % 60.8 % Tilt Class. With 0º 53.8 % 61.7 % Pan Class. With 15º 93.3 % 90.1 % Unknown Users LAAM separate LAAM grouped Pan Average Error 10.3 º 10.1 º Tilt Average Error 15.9 º 16.8 º Pan Class. With 0º 50.4 % 50 % Tilt Class. With 0º 43.9 % 44.5 % Pan Class. With 15º 88.1 % 88.7 % Table 1. Performance evaluation on known and unknown users. LAAM refers to linear auto-associative memories
7 7 Figure 2. Average error per pose for known and unknown users Concerning the pan angle, the average absolute error in pose is relatively stable with both techniques. Grouped and separated training accommodate intermediate tilt angles. We achieve a 0 degrees classification rate of 50.4% for pan angle and 21 % for tilt angles with the separate technique. Using the grouped technique provides a
8 8 50% classification rate for pan angle and 45% for tilt angle. Pan angle can be correctly estimated with a precision of 15 degrees in 88% of cases. These results tend to show that using the together technique does not provide significantly improve results. Examples can be seen in Figure 4. Faces are not aligned in the Pointing 04 database. Normalizing face images provides small variations in alignment. Results show that our system can handle alignment problems. Computing a score for each memory allows us to discriminate face and non-face images. Head detection and pose estimation is done in a single process. The system runs at 15 images/secs using the separate technique, and 3 images/secs with the grouped technique. As humans estimated angles separately, we will use the separate learning for comparison with human performances. 5.3 Comparison to Human Performances We computed the same evaluation measures for humans. Results for calibrated (C) and non-calibrated (NC) people are shown in Table 2. Global human average error for head pose estimation is 11.9 degrees in pan and 11 degrees in tilt. Creating two groups allows comparing the performances of our system on unknown users to the best human performances. We apply a Student test to compare the two populations. Calibrated people do not perform significantly much better in pan. However, the difference is significant in tilt angle. These results show that pan angle estimation appears to be natural for humans, whereas tilt angle estimation is not. This is due to the fact that people twist their head left and right more often than up and down during social interactions. In situations when people talk to each other, pan angle provides good cues on visual focus of attention [12], [19]. Head poses changes in tilt become meaningless. This is even more relevant when people are seated, because their head is roughly at the same height. People are more used to consider pose changes in pan. Seeing training images annotated do not improve much pan angle estimation but improves significantly tilt angle estimation. The best human performance is obtained by calibrated people. C NC LAAM Sep. U Pan Average Error 11.8 º 11.9 º 10.3 º Tilt Average Error 9.4 º 12.6 º 15.9 º Pan Class. with 0 º 40.7 % 42.4 % 50.4 % Tilt Class. With 0º 59 % 48 % 43.9 % Table 2. Human/Machine performance evaluation. C and NC stand for Calibrated and Non-Calibrated people
9 9 Figure 3. Human / System performance per pose Average error per pose for human subjects can be seen in Figure 3. For pan angle, we found that humans perform well for front and profile angles, but not for intermediate angles. The average error per pose in pan can be modelled by a Gaussian function centered at 45 degrees. Minimum error can be found at 0 degrees, which corresponds to front pose. Furthermore, during our experiment, we observe that most people did not use intermediate angles such as 30, 45 and 60 degrees. These results suggest that the human brain uses front and profile as key poses, as suggested in [17]. Concerning tilt angle, humans performs better for top angles than for bottom angles. The minimum error can be found at +90 degrees, whereas the maximum error is at -90 degrees. This can be due to the fact that when a face is nodding downward, hair dominates a large surface of the apparent face, providing more information about side to side angle. With an average error of 10.3 degrees and a precise classification rate of 50.4%, our method performs significantly better than humans at estimating pan angle (11.9 degrees). The standard deviation of the average error per pose is low for the system and high for humans. The system achieves roughly the same precision for front and profile, and higher precision for intermediate poses. With an average error of 11 degrees, humans perform better in tilt angle. Our method performs well for top poses. This is due to the fact that hair becomes more visible in the image and the face appearance between people changes more when looking down. On the other hand, such changes are less visible for up poses. Face region normalization also introduces a problem. The height of the neck changes between people. This provides high variations on face imagettes and can disrupt tilt angle estimation.
10 10 Figure 4. Pan angle estimation on example images 6. Conclusion We have proposed a new method to estimate head pose on unconstrained low resolution images. Face image is normalized in scale and slant into an imagette by a robust face detector. Face imagettes containing the same head pose are learned with the Widrow-Hoff correction rule to obtain a linear auto-associative memory. To estimate head pose, we compare source and reconstructed images using their cosine. A simple winner-takes-all process is applied to select the head pose which prototype gives the best match. We achieved a precision of 10.3 degrees in pan and 15.9 degrees in tilt only on unknown subjects on the Pointing 04 Head Pose Image database. Learning pan and tilt together does not provide significantly better results. Our method provides good results on very low resolution face images and can handle wide movements, which is particularly adapted to wide-angle or panoramic cameras setups. The system generalizes well to unknown users, is robust to alignment and runs at 15 frames/secs. We measured human performance on head pose estimation using the same data set. Our system performs significantly better than humans in pan, especially with intermediate angles. Humans perform better in tilt. Results of our system may be improved by fitting an ellipse to delimit more precisely the face. Our head pose estimation system can be adapted to video sequences for situations such as human interaction modelling, video surveillance and intelligent environments. By knowing a coarse estimate of the current head pose, the temporal context can help to limit head pose search only to neighbour poses. The use of head prototypes reduces significantly the computational time in video sequences. 7. References [1] A.H. Gee, R. Cipolla, Non-intrusive gaze tracking for human-computer interaction, Mechatronics and Machine Vision in Practise, pp , [2] R. Stiefelhagen, J. Yang, A. Waibel, "Tracking Eyes and Monitoring Eye Gaze, Workshop on Perceptual User Interfaces, pp , Banff, Canada, October 1997.
11 11 [3] A. Azarbayejani, T. Starner, B. Horowitz, A. Pentland, "Visually Controlled Graphics," IEEE Transactions on PAMI 15(6) 1993, pp [4] N. Gourier, D. Hall, J. Crowley, "Estimating Face Orientation using Robust Detection of Salient Facial Features," Pointing 2004, ICPR, Visual Observation of Deictic Gestures, Cambridge, UK. [5] J. Wu, J.M. Pedersen, D. Putthividhya, D. Norgaard, M.M. Trivedi, "A Two- Level Pose Estimation Framework Using Majority Voting of Gabor Wavelets and Bunch Graph Analysis," Pointing 2004, ICPR, Visual Observation of Deictic Gestures, Cambridge, UK. [6] Q. Chen, H. Wu, T. Fukumoto, M. Yachida, "3D Head Pose Estimation without Feature Tracking," AFGR, April 16/1998 Nara, Japan. pp [7] R. Brunelli, "Estimation of Pose and Illuminant Direction for Face Processing," Proceedings of IVC(15), No. 10, October 1997, pp [8] P. Yao, G. Evans, A. Calway, "Using Affine Correspondance to Estimate 3-D Facial Pose," 8th ICIP 2001, Thessaloniki, Greece, pp [9] S. McKenna, S. Gong, "Real-time face pose estimation," International Journal on Real Time Imaging, Special Issue on Real-time Visual Monitoring and Inspection, volume 4: pp , [10] J. Ng, S. Gong, "Multi-view Face Detection and Pose Estimation using a Composite Support Vector Machine across the View Sphere," International Workshop on Recognition, Analysis, and Tracking of Faces and Gestures in Real- Time Systems, p , Corfu, Greece, September [11] B. Schiele, A. Waibel, "Gaze tracking based on face-color," Workshop on Automatic Face and Gesture Recognition, pages , Zurich, June 26-28, [12] R. Stiefelhagen, "Estimating Head Pose with Neural Networks - Results on the Pointing04 ICPR Workshop Evaluation Data," Pointing 2004, ICPR, Visual Observation of Deictic Gestures, Cambridge, UK. [13] N. Gourier, J. Letessier, "The Pointing 04 Data Sets," Pointing 2004, ICPR, Visual Observation of Deictic Gestures, Cambridge, UK. [14] D. Valentin, H. Abdi, A. O'Toole, "Categorization and identification of human face images by neural networks: A review of linear auto-associator and principal component approaches," Journal of Biological Systems 2, pp , [15] K. Schwerdt, J. Crowley, "Robust face Tracking using Color," International Conference on Automatic face and Gesture Recognition pp , [16] G.J. Klinker, S.A. Shafer, T. Kanade, "A Physical Approach to Color Image Understanding," International Journal on Computer Vision [17] H. Abdi, D. Valentin, "Modeles Neuronaux, Connectionistes et Numeriques de la Reconnaissance des Visages," Psychologie Francaise, 39(4), pp , 1994.
12 12 [18] D. Kersten, N.F. Troje, H.H. Bülthoff, "Phenomenal competition for poses of the human head," Perception, 25 (1996), pp [19] B. Steinzor, The spatial factor in face to face discussions, Journal of Abnormal and Social Psychology 1950 (45), pp [20] H.H. Bülthoff, S.Y. Edelmann, M.J. Tarr, How are three-dimensional objects represented in the brain?, Cerebral Cortex 1995 (5) 3, pp [21] Seeing Machines Company. FaceLAB4, [22] H. Abdi, D. Valentin, "Modeles Neuronaux, Connectionistes et Numeriques de la Reconnaissance des Visages," Psychologie Francaise, 39(4), pp , 1994.
Taking Inverse Graphics Seriously
CSC2535: 2013 Advanced Machine Learning Taking Inverse Graphics Seriously Geoffrey Hinton Department of Computer Science University of Toronto The representation used by the neural nets that work best
More informationFace Model Fitting on Low Resolution Images
Face Model Fitting on Low Resolution Images Xiaoming Liu Peter H. Tu Frederick W. Wheeler Visualization and Computer Vision Lab General Electric Global Research Center Niskayuna, NY, 1239, USA {liux,tu,wheeler}@research.ge.com
More informationAccess control: adaptation and real time implantation of a face recognition method.
Access control: adaptation and real time implantation of a face recognition method. J. Mitéran, J.P. Zimmer, F. Yang, M. Paindavoine. Laboratory Le2i, University of Burgundy, Aile des Sciences de l'ingénieur,
More informationA Learning Based Method for Super-Resolution of Low Resolution Images
A Learning Based Method for Super-Resolution of Low Resolution Images Emre Ugur June 1, 2004 emre.ugur@ceng.metu.edu.tr Abstract The main objective of this project is the study of a learning based method
More informationFace Locating and Tracking for Human{Computer Interaction. Carnegie Mellon University. Pittsburgh, PA 15213
Face Locating and Tracking for Human{Computer Interaction Martin Hunke Alex Waibel School of Computer Science Carnegie Mellon University Pittsburgh, PA 15213 Abstract Eective Human-to-Human communication
More informationAnalecta Vol. 8, No. 2 ISSN 2064-7964
EXPERIMENTAL APPLICATIONS OF ARTIFICIAL NEURAL NETWORKS IN ENGINEERING PROCESSING SYSTEM S. Dadvandipour Institute of Information Engineering, University of Miskolc, Egyetemváros, 3515, Miskolc, Hungary,
More informationTemplate-based Eye and Mouth Detection for 3D Video Conferencing
Template-based Eye and Mouth Detection for 3D Video Conferencing Jürgen Rurainsky and Peter Eisert Fraunhofer Institute for Telecommunications - Heinrich-Hertz-Institute, Image Processing Department, Einsteinufer
More informationA PHOTOGRAMMETRIC APPRAOCH FOR AUTOMATIC TRAFFIC ASSESSMENT USING CONVENTIONAL CCTV CAMERA
A PHOTOGRAMMETRIC APPRAOCH FOR AUTOMATIC TRAFFIC ASSESSMENT USING CONVENTIONAL CCTV CAMERA N. Zarrinpanjeh a, F. Dadrassjavan b, H. Fattahi c * a Islamic Azad University of Qazvin - nzarrin@qiau.ac.ir
More informationThe Role of Size Normalization on the Recognition Rate of Handwritten Numerals
The Role of Size Normalization on the Recognition Rate of Handwritten Numerals Chun Lei He, Ping Zhang, Jianxiong Dong, Ching Y. Suen, Tien D. Bui Centre for Pattern Recognition and Machine Intelligence,
More informationA Real Time Hand Tracking System for Interactive Applications
A Real Time Hand Tracking System for Interactive Applications Siddharth Swarup Rautaray Indian Institute of Information Technology Allahabad ABSTRACT In vision based hand tracking systems color plays an
More informationCS231M Project Report - Automated Real-Time Face Tracking and Blending
CS231M Project Report - Automated Real-Time Face Tracking and Blending Steven Lee, slee2010@stanford.edu June 6, 2015 1 Introduction Summary statement: The goal of this project is to create an Android
More informationNormalisation of 3D Face Data
Normalisation of 3D Face Data Chris McCool, George Mamic, Clinton Fookes and Sridha Sridharan Image and Video Research Laboratory Queensland University of Technology, 2 George Street, Brisbane, Australia,
More informationFace detection is a process of localizing and extracting the face region from the
Chapter 4 FACE NORMALIZATION 4.1 INTRODUCTION Face detection is a process of localizing and extracting the face region from the background. The detected face varies in rotation, brightness, size, etc.
More informationAutomatic parameter regulation for a tracking system with an auto-critical function
Automatic parameter regulation for a tracking system with an auto-critical function Daniela Hall INRIA Rhône-Alpes, St. Ismier, France Email: Daniela.Hall@inrialpes.fr Abstract In this article we propose
More informationVISUAL RECOGNITION OF HAND POSTURES FOR INTERACTING WITH VIRTUAL ENVIRONMENTS
VISUAL RECOGNITION OF HAND POSTURES FOR INTERACTING WITH VIRTUAL ENVIRONMENTS Radu Daniel VATAVU Ştefan-Gheorghe PENTIUC "Stefan cel Mare" University of Suceava str.universitatii nr.13, RO-720229 Suceava
More informationThe Scientific Data Mining Process
Chapter 4 The Scientific Data Mining Process When I use a word, Humpty Dumpty said, in rather a scornful tone, it means just what I choose it to mean neither more nor less. Lewis Carroll [87, p. 214] In
More informationModelling, Extraction and Description of Intrinsic Cues of High Resolution Satellite Images: Independent Component Analysis based approaches
Modelling, Extraction and Description of Intrinsic Cues of High Resolution Satellite Images: Independent Component Analysis based approaches PhD Thesis by Payam Birjandi Director: Prof. Mihai Datcu Problematic
More informationPHYSIOLOGICALLY-BASED DETECTION OF COMPUTER GENERATED FACES IN VIDEO
PHYSIOLOGICALLY-BASED DETECTION OF COMPUTER GENERATED FACES IN VIDEO V. Conotter, E. Bodnari, G. Boato H. Farid Department of Information Engineering and Computer Science University of Trento, Trento (ITALY)
More informationMaking Machines Understand Facial Motion & Expressions Like Humans Do
Making Machines Understand Facial Motion & Expressions Like Humans Do Ana C. Andrés del Valle & Jean-Luc Dugelay Multimedia Communications Dpt. Institut Eurécom 2229 route des Crêtes. BP 193. Sophia Antipolis.
More informationInteractive person re-identification in TV series
Interactive person re-identification in TV series Mika Fischer Hazım Kemal Ekenel Rainer Stiefelhagen CV:HCI lab, Karlsruhe Institute of Technology Adenauerring 2, 76131 Karlsruhe, Germany E-mail: {mika.fischer,ekenel,rainer.stiefelhagen}@kit.edu
More informationFace Recognition in Low-resolution Images by Using Local Zernike Moments
Proceedings of the International Conference on Machine Vision and Machine Learning Prague, Czech Republic, August14-15, 014 Paper No. 15 Face Recognition in Low-resolution Images by Using Local Zernie
More informationHuman Behavior Interpretation System and Modeling
Proceedings of the 2003 IEEE International Conference on Robotics & Automation Taipei, Taiwan, September 14-19, 2003 Human Behavior Interpretation System based on View and Motion-based Aspect Models Masayuki
More informationPHOTOGRAMMETRIC TECHNIQUES FOR MEASUREMENTS IN WOODWORKING INDUSTRY
PHOTOGRAMMETRIC TECHNIQUES FOR MEASUREMENTS IN WOODWORKING INDUSTRY V. Knyaz a, *, Yu. Visilter, S. Zheltov a State Research Institute for Aviation System (GosNIIAS), 7, Victorenko str., Moscow, Russia
More informationTracking in flussi video 3D. Ing. Samuele Salti
Seminari XXIII ciclo Tracking in flussi video 3D Ing. Tutors: Prof. Tullio Salmon Cinotti Prof. Luigi Di Stefano The Tracking problem Detection Object model, Track initiation, Track termination, Tracking
More informationSubspace Analysis and Optimization for AAM Based Face Alignment
Subspace Analysis and Optimization for AAM Based Face Alignment Ming Zhao Chun Chen College of Computer Science Zhejiang University Hangzhou, 310027, P.R.China zhaoming1999@zju.edu.cn Stan Z. Li Microsoft
More informationFace Recognition For Remote Database Backup System
Face Recognition For Remote Database Backup System Aniza Mohamed Din, Faudziah Ahmad, Mohamad Farhan Mohamad Mohsin, Ku Ruhana Ku-Mahamud, Mustafa Mufawak Theab 2 Graduate Department of Computer Science,UUM
More informationRobust Real-Time Face Detection
Robust Real-Time Face Detection International Journal of Computer Vision 57(2), 137 154, 2004 Paul Viola, Michael Jones 授 課 教 授 : 林 信 志 博 士 報 告 者 : 林 宸 宇 報 告 日 期 :96.12.18 Outline Introduction The Boost
More informationFACE RECOGNITION BASED ATTENDANCE MARKING SYSTEM
Available Online at www.ijcsmc.com International Journal of Computer Science and Mobile Computing A Monthly Journal of Computer Science and Information Technology IJCSMC, Vol. 3, Issue. 2, February 2014,
More informationTracking Moving Objects In Video Sequences Yiwei Wang, Robert E. Van Dyck, and John F. Doherty Department of Electrical Engineering The Pennsylvania State University University Park, PA16802 Abstract{Object
More informationMonitoring Head/Eye Motion for Driver Alertness with One Camera
Monitoring Head/Eye Motion for Driver Alertness with One Camera Paul Smith, Mubarak Shah, and N. da Vitoria Lobo Computer Science, University of Central Florida, Orlando, FL 32816 rps43158,shah,niels @cs.ucf.edu
More informationTracking Groups of Pedestrians in Video Sequences
Tracking Groups of Pedestrians in Video Sequences Jorge S. Marques Pedro M. Jorge Arnaldo J. Abrantes J. M. Lemos IST / ISR ISEL / IST ISEL INESC-ID / IST Lisbon, Portugal Lisbon, Portugal Lisbon, Portugal
More informationNovelty Detection in image recognition using IRF Neural Networks properties
Novelty Detection in image recognition using IRF Neural Networks properties Philippe Smagghe, Jean-Luc Buessler, Jean-Philippe Urban Université de Haute-Alsace MIPS 4, rue des Frères Lumière, 68093 Mulhouse,
More informationVirtual Data Gloves : Interacting with Virtual Environments through Computer Vision
Virtual Data Gloves : Interacting with Virtual Environments through Computer Vision Richard Bowden (1), Tony Heap(2), Craig Hart(2) (1) Dept of M & ES (2) School of Computer Studies Brunel University University
More informationSpeed Performance Improvement of Vehicle Blob Tracking System
Speed Performance Improvement of Vehicle Blob Tracking System Sung Chun Lee and Ram Nevatia University of Southern California, Los Angeles, CA 90089, USA sungchun@usc.edu, nevatia@usc.edu Abstract. A speed
More informationProjection Center Calibration for a Co-located Projector Camera System
Projection Center Calibration for a Co-located Camera System Toshiyuki Amano Department of Computer and Communication Science Faculty of Systems Engineering, Wakayama University Sakaedani 930, Wakayama,
More informationQuantifying Spatial Presence. Summary
Quantifying Spatial Presence Cedar Riener and Dennis Proffitt Department of Psychology, University of Virginia Keywords: spatial presence, illusions, visual perception Summary The human visual system uses
More informationLOCAL SURFACE PATCH BASED TIME ATTENDANCE SYSTEM USING FACE. indhubatchvsa@gmail.com
LOCAL SURFACE PATCH BASED TIME ATTENDANCE SYSTEM USING FACE 1 S.Manikandan, 2 S.Abirami, 2 R.Indumathi, 2 R.Nandhini, 2 T.Nanthini 1 Assistant Professor, VSA group of institution, Salem. 2 BE(ECE), VSA
More informationObject Recognition and Template Matching
Object Recognition and Template Matching Template Matching A template is a small image (sub-image) The goal is to find occurrences of this template in a larger image That is, you want to find matches of
More informationReconstructing 3D Pose and Motion from a Single Camera View
Reconstructing 3D Pose and Motion from a Single Camera View R Bowden, T A Mitchell and M Sarhadi Brunel University, Uxbridge Middlesex UB8 3PH richard.bowden@brunel.ac.uk Abstract This paper presents a
More informationExtracting a Good Quality Frontal Face Images from Low Resolution Video Sequences
Extracting a Good Quality Frontal Face Images from Low Resolution Video Sequences Pritam P. Patil 1, Prof. M.V. Phatak 2 1 ME.Comp, 2 Asst.Professor, MIT, Pune Abstract The face is one of the important
More informationIllumination, Expression and Occlusion Invariant Pose-Adaptive Face Recognition System for Real- Time Applications
Illumination, Expression and Occlusion Invariant Pose-Adaptive Face Recognition System for Real- Time Applications Shireesha Chintalapati #1, M. V. Raghunadh *2 Department of E and CE NIT Warangal, Andhra
More informationShape Measurement of a Sewer Pipe. Using a Mobile Robot with Computer Vision
International Journal of Advanced Robotic Systems ARTICLE Shape Measurement of a Sewer Pipe Using a Mobile Robot with Computer Vision Regular Paper Kikuhito Kawasue 1,* and Takayuki Komatsu 1 1 Department
More informationOBJECT TRACKING USING LOG-POLAR TRANSFORMATION
OBJECT TRACKING USING LOG-POLAR TRANSFORMATION A Thesis Submitted to the Gradual Faculty of the Louisiana State University and Agricultural and Mechanical College in partial fulfillment of the requirements
More informationHow To Fix Out Of Focus And Blur Images With A Dynamic Template Matching Algorithm
IJSTE - International Journal of Science Technology & Engineering Volume 1 Issue 10 April 2015 ISSN (online): 2349-784X Image Estimation Algorithm for Out of Focus and Blur Images to Retrieve the Barcode
More informationReal-Time Tracking of Pedestrians and Vehicles
Real-Time Tracking of Pedestrians and Vehicles N.T. Siebel and S.J. Maybank. Computational Vision Group Department of Computer Science The University of Reading Reading RG6 6AY, England Abstract We present
More informationHANDS-FREE PC CONTROL CONTROLLING OF MOUSE CURSOR USING EYE MOVEMENT
International Journal of Scientific and Research Publications, Volume 2, Issue 4, April 2012 1 HANDS-FREE PC CONTROL CONTROLLING OF MOUSE CURSOR USING EYE MOVEMENT Akhil Gupta, Akash Rathi, Dr. Y. Radhika
More informationMean-Shift Tracking with Random Sampling
1 Mean-Shift Tracking with Random Sampling Alex Po Leung, Shaogang Gong Department of Computer Science Queen Mary, University of London, London, E1 4NS Abstract In this work, boosting the efficiency of
More information3D Model based Object Class Detection in An Arbitrary View
3D Model based Object Class Detection in An Arbitrary View Pingkun Yan, Saad M. Khan, Mubarak Shah School of Electrical Engineering and Computer Science University of Central Florida http://www.eecs.ucf.edu/
More informationDRIVER HEAD POSE AND VIEW ESTIMATION WITH SINGLE OMNIDIRECTIONAL VIDEO STREAM
DRIVER HEAD POSE AND VIEW ESTIMATION WITH SINGLE OMNIDIRECTIONAL VIDEO STREAM Kohsia S. Huang and Mohan M. Trivedi Computer Vision and Robotics Research (CVRR) Laboratory University of California, San
More informationAN IMPROVED DOUBLE CODING LOCAL BINARY PATTERN ALGORITHM FOR FACE RECOGNITION
AN IMPROVED DOUBLE CODING LOCAL BINARY PATTERN ALGORITHM FOR FACE RECOGNITION Saurabh Asija 1, Rakesh Singh 2 1 Research Scholar (Computer Engineering Department), Punjabi University, Patiala. 2 Asst.
More informationJournal of Industrial Engineering Research. Adaptive sequence of Key Pose Detection for Human Action Recognition
IWNEST PUBLISHER Journal of Industrial Engineering Research (ISSN: 2077-4559) Journal home page: http://www.iwnest.com/aace/ Adaptive sequence of Key Pose Detection for Human Action Recognition 1 T. Sindhu
More informationThe Implementation of Face Security for Authentication Implemented on Mobile Phone
The Implementation of Face Security for Authentication Implemented on Mobile Phone Emir Kremić *, Abdulhamit Subaşi * * Faculty of Engineering and Information Technology, International Burch University,
More informationInternational Journal of Advanced Information in Arts, Science & Management Vol.2, No.2, December 2014
Efficient Attendance Management System Using Face Detection and Recognition Arun.A.V, Bhatath.S, Chethan.N, Manmohan.C.M, Hamsaveni M Department of Computer Science and Engineering, Vidya Vardhaka College
More informationFeature Tracking and Optical Flow
02/09/12 Feature Tracking and Optical Flow Computer Vision CS 543 / ECE 549 University of Illinois Derek Hoiem Many slides adapted from Lana Lazebnik, Silvio Saverse, who in turn adapted slides from Steve
More informationVisual-based ID Verification by Signature Tracking
Visual-based ID Verification by Signature Tracking Mario E. Munich and Pietro Perona California Institute of Technology www.vision.caltech.edu/mariomu Outline Biometric ID Visual Signature Acquisition
More informationUNIVERSITY OF BOLTON SCHOOL OF ENGINEERING MS SYSTEMS ENGINEERING AND ENGINEERING MANAGEMENT SEMESTER 1 EXAMINATION 2015/2016 INTELLIGENT SYSTEMS
TW72 UNIVERSITY OF BOLTON SCHOOL OF ENGINEERING MS SYSTEMS ENGINEERING AND ENGINEERING MANAGEMENT SEMESTER 1 EXAMINATION 2015/2016 INTELLIGENT SYSTEMS MODULE NO: EEM7010 Date: Monday 11 th January 2016
More informationPotential of face area data for predicting sharpness of natural images
Potential of face area data for predicting sharpness of natural images Mikko Nuutinen a, Olli Orenius b, Timo Säämänen b, Pirkko Oittinen a a Dept. of Media Technology, Aalto University School of Science
More informationPart-Based Recognition
Part-Based Recognition Benedict Brown CS597D, Fall 2003 Princeton University CS 597D, Part-Based Recognition p. 1/32 Introduction Many objects are made up of parts It s presumably easier to identify simple
More informationVEHICLE LOCALISATION AND CLASSIFICATION IN URBAN CCTV STREAMS
VEHICLE LOCALISATION AND CLASSIFICATION IN URBAN CCTV STREAMS Norbert Buch 1, Mark Cracknell 2, James Orwell 1 and Sergio A. Velastin 1 1. Kingston University, Penrhyn Road, Kingston upon Thames, KT1 2EE,
More informationAn Approach for Utility Pole Recognition in Real Conditions
6th Pacific-Rim Symposium on Image and Video Technology 1st PSIVT Workshop on Quality Assessment and Control by Image and Video Analysis An Approach for Utility Pole Recognition in Real Conditions Barranco
More informationINTRODUCTION TO MACHINE LEARNING 3RD EDITION
ETHEM ALPAYDIN The MIT Press, 2014 Lecture Slides for INTRODUCTION TO MACHINE LEARNING 3RD EDITION alpaydin@boun.edu.tr http://www.cmpe.boun.edu.tr/~ethem/i2ml3e CHAPTER 1: INTRODUCTION Big Data 3 Widespread
More informationDigital image processing
746A27 Remote Sensing and GIS Lecture 4 Digital image processing Chandan Roy Guest Lecturer Department of Computer and Information Science Linköping University Digital Image Processing Most of the common
More informationApplication Example: Quality Control of Injection-Molded Parts
Application Example: Quality Control of Injection-Molded Parts Measuring Systems: ATOS Keywords: Optical 3D measuring technology, mold try-out, first article inspection, form and dimension inspection of
More informationA Short Introduction to Computer Graphics
A Short Introduction to Computer Graphics Frédo Durand MIT Laboratory for Computer Science 1 Introduction Chapter I: Basics Although computer graphics is a vast field that encompasses almost any graphical
More informationPoker Vision: Playing Cards and Chips Identification based on Image Processing
Poker Vision: Playing Cards and Chips Identification based on Image Processing Paulo Martins 1, Luís Paulo Reis 2, and Luís Teófilo 2 1 DEEC Electrical Engineering Department 2 LIACC Artificial Intelligence
More informationECE 533 Project Report Ashish Dhawan Aditi R. Ganesan
Handwritten Signature Verification ECE 533 Project Report by Ashish Dhawan Aditi R. Ganesan Contents 1. Abstract 3. 2. Introduction 4. 3. Approach 6. 4. Pre-processing 8. 5. Feature Extraction 9. 6. Verification
More informationGeometric Camera Parameters
Geometric Camera Parameters What assumptions have we made so far? -All equations we have derived for far are written in the camera reference frames. -These equations are valid only when: () all distances
More informationLimitations of Human Vision. What is computer vision? What is computer vision (cont d)?
What is computer vision? Limitations of Human Vision Slide 1 Computer vision (image understanding) is a discipline that studies how to reconstruct, interpret and understand a 3D scene from its 2D images
More informationAn Active Head Tracking System for Distance Education and Videoconferencing Applications
An Active Head Tracking System for Distance Education and Videoconferencing Applications Sami Huttunen and Janne Heikkilä Machine Vision Group Infotech Oulu and Department of Electrical and Information
More informationHAND GESTURE BASEDOPERATINGSYSTEM CONTROL
HAND GESTURE BASEDOPERATINGSYSTEM CONTROL Garkal Bramhraj 1, palve Atul 2, Ghule Supriya 3, Misal sonali 4 1 Garkal Bramhraj mahadeo, 2 Palve Atule Vasant, 3 Ghule Supriya Shivram, 4 Misal Sonali Babasaheb,
More informationProduct Information. QUADRA-CHEK 3000 Evaluation Electronics For Metrological Applications
Product Information QUADRA-CHEK 3000 Evaluation Electronics For Metrological Applications April 2016 QUADRA-CHEK 3000 The evaluation electronics for intuitive 2-D measurement The QUADRA-CHEK 3000 evaluation
More informationRobust Real-Time Face Tracking and Gesture Recognition
Robust Real-Time Face Tracking and Gesture Recognition J. Heinzmann and A. Zelinsky Department of Systems Engineering Research School of Information Sciences and Engineering Australian National University
More informationDiscrimination of Gaze Directions Using Low-Level Eye Image Features
Discrimination of Gaze Directions Using Low-Level Eye Image Features Yanxia Zhang Lancaster University yazhang@lancasteracuk Andreas Bulling University of Cambridge & Lancaster University andreasbulling@acmorg
More informationIncremental PCA: An Alternative Approach for Novelty Detection
Incremental PCA: An Alternative Approach for Detection Hugo Vieira Neto and Ulrich Nehmzow Department of Computer Science University of Essex Wivenhoe Park Colchester CO4 3SQ {hvieir, udfn}@essex.ac.uk
More informationNeural Network based Vehicle Classification for Intelligent Traffic Control
Neural Network based Vehicle Classification for Intelligent Traffic Control Saeid Fazli 1, Shahram Mohammadi 2, Morteza Rahmani 3 1,2,3 Electrical Engineering Department, Zanjan University, Zanjan, IRAN
More informationEFFICIENT VEHICLE TRACKING AND CLASSIFICATION FOR AN AUTOMATED TRAFFIC SURVEILLANCE SYSTEM
EFFICIENT VEHICLE TRACKING AND CLASSIFICATION FOR AN AUTOMATED TRAFFIC SURVEILLANCE SYSTEM Amol Ambardekar, Mircea Nicolescu, and George Bebis Department of Computer Science and Engineering University
More informationUniversidad de Cantabria Departamento de Tecnología Electrónica, Ingeniería de Sistemas y Automática. Tesis Doctoral
Universidad de Cantabria Departamento de Tecnología Electrónica, Ingeniería de Sistemas y Automática Tesis Doctoral CONTRIBUCIONES AL ALINEAMIENTO DE NUBES DE PUNTOS 3D PARA SU USO EN APLICACIONES DE CAPTURA
More informationVision-based Walking Parameter Estimation for Biped Locomotion Imitation
Vision-based Walking Parameter Estimation for Biped Locomotion Imitation Juan Pedro Bandera Rubio 1, Changjiu Zhou 2 and Francisco Sandoval Hernández 1 1 Dpto. Tecnología Electrónica, E.T.S.I. Telecomunicación
More informationThe Advantages of Using a Fixed Stereo Vision sensor
Proc. of International Conference on Industrial & Engineering Applications of Artificial Intelligence & Expert Systems (IEA/AIE), 2005 Real-Time People Localization and Tracking through Fixed Stereo Vision
More informationBernice E. Rogowitz and Holly E. Rushmeier IBM TJ Watson Research Center, P.O. Box 704, Yorktown Heights, NY USA
Are Image Quality Metrics Adequate to Evaluate the Quality of Geometric Objects? Bernice E. Rogowitz and Holly E. Rushmeier IBM TJ Watson Research Center, P.O. Box 704, Yorktown Heights, NY USA ABSTRACT
More informationClassifying Manipulation Primitives from Visual Data
Classifying Manipulation Primitives from Visual Data Sandy Huang and Dylan Hadfield-Menell Abstract One approach to learning from demonstrations in robotics is to make use of a classifier to predict if
More informationComponent Ordering in Independent Component Analysis Based on Data Power
Component Ordering in Independent Component Analysis Based on Data Power Anne Hendrikse Raymond Veldhuis University of Twente University of Twente Fac. EEMCS, Signals and Systems Group Fac. EEMCS, Signals
More informationVision based approach to human fall detection
Vision based approach to human fall detection Pooja Shukla, Arti Tiwari CSVTU University Chhattisgarh, poojashukla2410@gmail.com 9754102116 Abstract Day by the count of elderly people living alone at home
More informationAutomatic Calibration of an In-vehicle Gaze Tracking System Using Driver s Typical Gaze Behavior
Automatic Calibration of an In-vehicle Gaze Tracking System Using Driver s Typical Gaze Behavior Kenji Yamashiro, Daisuke Deguchi, Tomokazu Takahashi,2, Ichiro Ide, Hiroshi Murase, Kazunori Higuchi 3,
More informationSegmentation & Clustering
EECS 442 Computer vision Segmentation & Clustering Segmentation in human vision K-mean clustering Mean-shift Graph-cut Reading: Chapters 14 [FP] Some slides of this lectures are courtesy of prof F. Li,
More informationGLOVE-BASED GESTURE RECOGNITION SYSTEM
CLAWAR 2012 Proceedings of the Fifteenth International Conference on Climbing and Walking Robots and the Support Technologies for Mobile Machines, Baltimore, MD, USA, 23 26 July 2012 747 GLOVE-BASED GESTURE
More informationHow To Compress Video For Real Time Transmission
University of Edinburgh College of Science and Engineering School of Informatics Informatics Research Proposal supervised by Dr. Sethu Vijayakumar Optimized bandwidth usage for real-time remote surveillance
More informationSupporting Online Material for
www.sciencemag.org/cgi/content/full/313/5786/504/dc1 Supporting Online Material for Reducing the Dimensionality of Data with Neural Networks G. E. Hinton* and R. R. Salakhutdinov *To whom correspondence
More informationA Prototype For Eye-Gaze Corrected
A Prototype For Eye-Gaze Corrected Video Chat on Graphics Hardware Maarten Dumont, Steven Maesen, Sammy Rogmans and Philippe Bekaert Introduction Traditional webcam video chat: No eye contact. No extensive
More informationPASSIVE DRIVER GAZE TRACKING WITH ACTIVE APPEARANCE MODELS
PASSIVE DRIVER GAZE TRACKING WITH ACTIVE APPEARANCE MODELS Takahiro Ishikawa Research Laboratories, DENSO CORPORATION Nisshin, Aichi, Japan Tel: +81 (561) 75-1616, Fax: +81 (561) 75-1193 Email: tishika@rlab.denso.co.jp
More informationAn Iterative Image Registration Technique with an Application to Stereo Vision
An Iterative Image Registration Technique with an Application to Stereo Vision Bruce D. Lucas Takeo Kanade Computer Science Department Carnegie-Mellon University Pittsburgh, Pennsylvania 15213 Abstract
More informationCommunicating Agents Architecture with Applications in Multimodal Human Computer Interaction
Communicating Agents Architecture with Applications in Multimodal Human Computer Interaction Maximilian Krüger, Achim Schäfer, Andreas Tewes, Rolf P. Würtz Institut für Neuroinformatik, Ruhr-Universität
More informationIndex Terms: Face Recognition, Face Detection, Monitoring, Attendance System, and System Access Control.
Modern Technique Of Lecture Attendance Using Face Recognition. Shreya Nallawar, Neha Giri, Neeraj Deshbhratar, Shamal Sane, Trupti Gautre, Avinash Bansod Bapurao Deshmukh College Of Engineering, Sewagram,
More informationHands Tracking from Frontal View for Vision-Based Gesture Recognition
Hands Tracking from Frontal View for Vision-Based Gesture Recognition Jörg Zieren, Nils Unger, and Suat Akyol Chair of Technical Computer Science, Ahornst. 55, Aachen University (RWTH), 52074 Aachen, Germany
More informationPerformance Comparison of Visual and Thermal Signatures for Face Recognition
Performance Comparison of Visual and Thermal Signatures for Face Recognition Besma Abidi The University of Tennessee The Biometric Consortium Conference 2003 September 22-24 OUTLINE Background Recognition
More informationFace Identification by Human and by Computer: Two Sides of the Same Coin, or Not? Tsuhan Chen tsuhan@cmu.edu
Face Identification by Human and by Computer: Two Sides of the Same Coin, or Not? Tsuhan Chen tsuhan@cmu.edu Carnegie Mellon University Pittsburgh, USA What do you see? 1 What do you see? 2 What do you
More informationPerception-based Design for Tele-presence
Perception-based Design for Tele-presence Santanu Chaudhury 1, Shantanu Ghosh 1,2, Amrita Basu 3, Brejesh Lall 1, Sumantra Dutta Roy 1, Lopamudra Choudhury 3, Prashanth R 1, Ashish Singh 1, and Amit Maniyar
More informationFast Image Mosaicing for Panoramic Face Recognition
14 JOURNAL OF MULTIMEDIA, VOL. 1, NO. 2, MAY 2006 Fast Image Mosaicing for Panoramic Face Recognition Fan Yang, Michel Paindavoine, Hervé Abdi*, Dominique Arnoult Laboratoire Le2i, Aile de l Ingénieur
More informationImage Normalization for Illumination Compensation in Facial Images
Image Normalization for Illumination Compensation in Facial Images by Martin D. Levine, Maulin R. Gandhi, Jisnu Bhattacharyya Department of Electrical & Computer Engineering & Center for Intelligent Machines
More informationAccurate and robust image superresolution by neural processing of local image representations
Accurate and robust image superresolution by neural processing of local image representations Carlos Miravet 1,2 and Francisco B. Rodríguez 1 1 Grupo de Neurocomputación Biológica (GNB), Escuela Politécnica
More information